Momentic Debuts Mo AI Agent to Automate Software Testing Without Scripts

Whenever code is generated — by a human or an AI agent — something needs to test the app for bugs before production. That is typically the job of a quality assurance engineer. The rapid pace of modern code creation means errors and issues can slip through unnoticed, and AI-driven build cycles have changed how developers work. A developer might oversee a coding agent, review its output, correct obvious errors and pass the code to testing. The tests themselves are code that must be written, maintained and verified.

“What is the next frontier of agentic quality? Our bet is that there’s no tests in the future,” co-founder and Chief Executive Wei-Wei Wu told SiliconANGLE. “It’s just agents verifying it themselves, based off of your instructions.”

There are two primary ways to do testing: tests that check code quality and tests that verify that the application works when it is running. Scripts for the latter click buttons, trigger drop-down menus and fill in fields, but humans are better because they operate the app where it will be used. AI agents can emulate that behavior, Momentic said. Testing an app for release can be simple at first: write enough tests to ensure buttons work, a user can reach the shopping cart, add carrots and check out. But as applications become more complex, more tests are needed, and it eventually becomes impossible to keep up. The result is a bloat of tests that developers must check, modify and fix every time the app changes; otherwise, the scripts become a liability.

“As long as you have scripts somewhere in your codebase, someone has to maintain it,” Wu said.

Momentic originally tried to address the maintenance burden of scripted browser tests by helping developers describe workflows in ordinary language, which a large language model can reduce the labor for. Wu said the company eventually ran into a deeper version of the same problem: even a much easier-to-maintain test artifact is still something somebody has to maintain. Mo removes the intermediate layer and allows QA agents to work from requirements and intent directly.

“With agents, the economics of scale can actually work,” Wu said. Instead of building and maintaining a second repository of scripts, a developer can point Mo at the app and ask it to test it, according to the company.

Mo is designed to look like other AI-coding tools developers use day to day, such as Claude Code. It has an agentic harness with a web interface and a chat box where a developer can enter what they need and a URL to get started. It also includes a command-line interface for automating tasks. Wu described a typical instruction this way: “Bug bash this app I just built. Here’s the URL. Here are some test credentials. Go ham.”

Once Mo takes the reins, it opens the app, spins up a swarm of agents and goes through the motions. Beyond a prompt telling the agent what to test or the intent to test, it can take a product requirements document, Jira issue, Linear ticket or Confluence document when a feature needs more context. Mo then uses the agentic swarm to attempt thousands of permutations and edge cases and records every interaction. When it completes, it gives the developer a blow-by-blow of what was tested, what failed, how to reproduce the bugs and video evidence of each one. There is no script to write and no test suite to maintain. Wu added that if developers want to turn these into tests, the fact that they are reproducible and fully described means developers can turn them into the script of their choosing.

Momentic said customers have been putting Mo through its paces over the past few weeks, including technology and service industry companies that use web and mobile apps such as Notion, Superpower, Iris, Committee for Children and Boundless.

“The scripts are not the end goal,” Wu said. “No one goes to work and is like, ‘Hey, I want to write 10,000 Momentic scripts or Playwright scripts or Selenium scripts.’ What they want is no bugs.”

Editor's Summary

Momentic has launched Mo, an AI agent that aims to test running software without requiring teams to write or maintain scripts. The product works from natural-language prompts and development documents, records failures and returns reproduction details, and has been tested by customers including Notion. Its success would depend on whether agent-based testing can replace scripted QA at scale.